Upstream bug: https://code.google.com/p/chromium/issues/detail?id=416856 Reproducible: Always Steps to Reproduce: 1. go to hangouts in chromium 2. click the share screen button Actual Results: Nothing happens Expected Results: The chooseDesktopMedia popup appears as it does in chrome There's a fix suggested in the bug [1], and I filed a patch for this (just for the most recent ebuild) at [2]. [1]: https://code.google.com/p/chromium/issues/detail?id=416856#c53 [2]: https://github.com/gentoo/gentoo/pull/166
Created attachment 414146 [details, diff] Patch version of https://github.com/gentoo/gentoo/pull/166 $ diff -u chromium-47.0.2526.6{,-r1}.ebuild --- chromium-47.0.2526.6.ebuild 2015-10-08 11:40:39.217708149 -0700 +++ chromium-47.0.2526.6-r1.ebuild 2015-10-08 11:43:06.564715487 -0700 @@ -19,7 +19,7 @@ LICENSE="BSD hotwording? ( no-source-code )" SLOT="0" KEYWORDS="~amd64 ~x86" -IUSE="cups gnome gnome-keyring gtk3 hidpi hotwording kerberos neon pic +proprietary-codecs pulseaudio selinux +system-ffmpeg +tcmalloc widevine" +IUSE="cups gnome gnome-keyring gtk3 +hangout hidpi hotwording kerberos neon pic +proprietary-codecs pulseaudio selinux +system-ffmpeg +tcmalloc widevine" RESTRICT="!system-ffmpeg? ( proprietary-codecs? ( bindist ) )" # Native Client binaries are compiled with different set of flags, bug #452066. @@ -378,6 +378,7 @@ $(gyp_use gnome-keyring use_gnome_keyring) $(gyp_use gnome-keyring linux_link_gnome_keyring) $(gyp_use gtk3) + $(gyp_use hangout enable_hangout_services_extension) $(gyp_use hidpi enable_hidpi) $(gyp_use hotwording enable_hotwording) $(gyp_use kerberos)
Does this patch fix it? The patches I gave in the Google Code bug report have not given me the expected results. I did not try directly modifying the ebuild but it seems it would do the same thing. Also as I stated there, I think (but I am not sure) this may have licensing implications, such that distributing a Chromium build with this enabled 'puts you in a legal problem with Google' just as with custom builds of Firefox that are branded as such. A postinst message about this should probably be added, and this feature should probably have a USE flag.
(In reply to Andrew Udvare from comment #2) > Does this patch fix it? The patches I gave in the Google Code bug report > have not given me the expected results. I did not try directly modifying the > ebuild but it seems it would do the same thing. I didn't try your patches, but I did run this patch and it worked. Although I ran it on the amd64-stable www-client/chromium-45.0.2454.101, I haven't actually tested on 47.0.2526.6. > Also as I stated there, I think (but I am not sure) this may have licensing > implications, such that distributing a Chromium build with this enabled > 'puts you in a legal problem with Google' just as with custom builds of > Firefox that are branded as such. A postinst message about this should > probably be added, and this feature should probably have a USE flag. I did put it behind a ‘hangout’ USE flag (see the smaller diff in [1]), but I enabled that flag by default. It may be better (until we get clarity on the licensing) to require -hangout if bindist is set. [1]: https://bugs.gentoo.org/show_bug.cgi?id=562590#c1
Currently trying your patch with version 46.0.2490.42 and will report back.
This works great. Tested 46.0.2490.42. I'm guessing this is licensing-related since the common.gypi file disables the hangout extra stuff for builds that are not branded as Chrome and do not enforce enabling hangout extensions in the main configuration. It does not seem like there is a technical reason that the feature gets disabled as there is no mention of Linux about this in common.gypi. > I did put it behind a ‘hangout’ USE flag (see the smaller diff in [1]), but I enabled that flag by default. It may be better (until we get clarity on the licensing) to require -hangout if bindist is set. -hangout required for bindist sounds like a good idea, to stay on the safe side.
I've applied the fixes to 48 and 47, see https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=13157f3d1aa2c55bfed1bc3bb28dbf036a184956 and https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=0966a024c5cb0277859829059cdd3d0c53cdc3a2 I plan to keep this bug open until the fixes make it to stable.
Thanks. I manually did that patch to 47 and it works.
If we're still concerned about licensing issues, we should add something to RESTRICT. There's already a RESTRICT setting in the chromium ebuilds, so we probably just want to append ‘hangouts? ( bindist )’ to those, and maybe make the default ‘-hangouts’.
I dont even know how I can use this stuff to fix my chromium. I have searched out and found this thread and now am really excited i may get this fixed. Im really new to linux. I dont even know what gentoo is or how to implement this fix. Any advice?
(In reply to W. Trevor King from comment #8) > If we're still concerned about licensing issues, we should add something to > RESTRICT. There's already a RESTRICT setting in the chromium ebuilds, so we > probably just want to append ‘hangouts? ( bindist )’ to those, and maybe > make the default ‘-hangouts’. It is still very unclear; have to ask upstream I guess. The change that allows all builds to specify Hangouts extension does not say anything about licensing. It would seem like that is what it is related to, because if gyp receives no flag about hangouts and the build type is Chrome, then hangouts is enabled, but otherwise not. This is why I mentioned licensing early on. Chrome+Hangouts = ok, no distributing except by Google because branding is Chrome Chromium+Hangouts works technically, but distribution license required (probably for OEMs making custom Chromium builds); end users here cannot distribute. https://chromium.googlesource.com/chromium/src/+/754a2004d20eb281a79e72bdd9dd59692ec4e7cc%5E!/
(In reply to Andrew Udvare from comment #10) > https://chromium.googlesource.com/chromium/src/+/ > 754a2004d20eb281a79e72bdd9dd59692ec4e7cc%5E!/ The linked review page [1] doesn't say anything about licensing either. [1]: https://codereview.chromium.org/163583004
(In reply to W. Trevor King from comment #11) > The linked review page [1] doesn't say anything about licensing either. > > [1]: https://codereview.chromium.org/163583004 And there's nothing about licensing in the issue linked from the review page [1]. [1]: https://code.google.com/p/chromium/issues/detail?id=343567
chromium-47 is now stable, closing